<?xml version="1.0" encoding="utf-8" standalone="yes"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:content="http://purl.org/rss/1.0/modules/content/">
  <channel>
    <title>Airesearch on Tenu Tech Brief</title>
    <link>https://cluster-site.onrender.com/tags/airesearch/</link>
    <description>Recent content in Airesearch on Tenu Tech Brief</description>
    <generator>Hugo -- 0.146.0</generator>
    <language>en-us</language>
    <lastBuildDate>Tue, 24 Feb 2026 06:06:03 +0000</lastBuildDate>
    <atom:link href="https://cluster-site.onrender.com/tags/airesearch/index.xml" rel="self" type="application/rss+xml" />
    <item>
      <title>BiScale: Energy-Efficient Disaggregated LLM Serving via Phase-Aware Placement and DVFS</title>
      <link>https://cluster-site.onrender.com/posts/biscale-energy-efficient-disaggregated-llm-serving-via-phase-aware-placement-and-dvfs/</link>
      <pubDate>Tue, 24 Feb 2026 05:00:00 +0000</pubDate>
      <guid>https://cluster-site.onrender.com/posts/biscale-energy-efficient-disaggregated-llm-serving-via-phase-aware-placement-and-dvfs/</guid>
      <description>• Prefill/decode disaggregation improves latency-throughput tradeoff for large language model serving. • Energy consumption remains high; autoscaling is too coarse-grained for rapi</description>
    </item>
    <item>
      <title>FineRef: Fine-Grained Error Reflection and Correction for Long-Form Generation with Citations</title>
      <link>https://cluster-site.onrender.com/posts/fineref-fine-grained-error-reflection-and-correction-for-long-form-generation-with-citations/</link>
      <pubDate>Tue, 24 Feb 2026 05:00:00 +0000</pubDate>
      <guid>https://cluster-site.onrender.com/posts/fineref-fine-grained-error-reflection-and-correction-for-long-form-generation-with-citations/</guid>
      <description>• FineRef introduces fine-grained error reflection for citation mismatch and irrelevance in long‑form LLM generation. • Two‑stage training: supervised fine‑tuning with attempt‑refl</description>
    </item>
    <item>
      <title>Quantifying construct validity in large language model evaluations</title>
      <link>https://cluster-site.onrender.com/posts/quantifying-construct-validity-in-large-language-model-evaluations/</link>
      <pubDate>Wed, 18 Feb 2026 05:00:00 +0000</pubDate>
      <guid>https://cluster-site.onrender.com/posts/quantifying-construct-validity-in-large-language-model-evaluations/</guid>
      <description>• LLM benchmarks often misrepresent true model capabilities due to contamination and annotator errors. • Construct validity is essential to ensure benchmarks truly measure desired</description>
    </item>
  </channel>
</rss>
